CIFECENTER FOR INTEGRATED FACILITY ENGINEERING

CIFECENTER FOR INTEGRATED FACILITY ENGINEERING Virtual Design and Construction: Themes, Case Studies and Implementation Suggestions By John Kunz & Martin Fischer CIFE Working Paper #097 Version 14; January 2012 STANFORD UNIVERSITY

COPYRIGHT 2012 BY Center for Integrated Facility Engineering If you would like to contact the authors, please write to: c/o CIFE, Civil and Environmental Engineering Dept., Stanford University The Jerry Yang & Akiko Yamazaki Environment & Energy Building 473 Via Ortega, Mail Code: 4020 Stanford, CA 94305-4020

Virtual Design and Construction: Themes, Case Studies and Implementation Suggestions John Kunz and Martin Fischer CIFE, Stanford University Abstract Virtual Design and Construction (VDC 1 ) is the use of integrated multi-disciplinary performance models of design-construction projects to support explicit and public business objectives. This paper describes the theory and methods of VDC, and it includes specific examples of models and precise objectives as well as detailed suggestions on how to implement VDC in practice. VDC models are virtual because they show computer-based descriptions of the project. The VDC project model emphasizes those aspects of the project that can be designed and managed, i.e., the product (typically a building or plant), the organization that will define, design, construct and operate it, and the process that the organization teams will follow. These models are logically integrated in the sense that they all can access shared data, and if a user highlights or changes an aspect of one, the integrated models can highlight or change the dependent aspects of related models. The models are multi-disciplinary in the sense that they represent the Architect, Engineering, Contractor (AEC) and Owner of the project, as well as relevant sub disciplines. The models are performance models in the sense that they predict some aspects of project performance, track many that are relevant, and can show predicted and measured performance in relationship to stated project performance objectives. Some companies now practice the first steps of VDC modeling, and they consistently find that they improve business performance by doing so. 1 Italics indicates that the glossary defines the italicized term 1/19/2012 1

Virtual Design and Construction:... 1 Themes, Case Studies and Implementation Suggestions... 1 Abstract... 1 Background... 2 VDC builds on traditional (20 th century) practice... 2 Early History... 3 Example... 4 VDC maturity model... 4 Themes... 6 VDC models are virtual... 6 VDC models represent the Product, Organization and Process (POP)... 9 POP models have different levels of detail... 14 Use Breakdown Structures to define generic POP models... 17 VDC product models show physical and abstract elements of a design... 22 4D animations visualize the product as it is built during the construction process... 23 Project models include the Organization and Process... 23 VDC supports public and explicit business objectives and metrics... 25 VDC Objectives Framework... 25 VDC emerges in stages... 32 Visualization shows the product, organization and process design... 33 Automation: automate some routine design and pre-fabricate to enable subassembly installation... 34 Integrated Concurrent Engineering (ICE) supports VDC... 34 VDC models support and require economic impact analysis... 38 Summary... 41 Discussion... 42 VDC strategy can enable companies to achieve significant breakthrough objectives.. 42 Different stakeholders have different responsibilities... 42 Stakeholders collaborate by sharing visualizations... 43 VDC enables better project management... 43 VDC Limitations... 43 Glossary... 44 Acknowledgements... 47 References... 47 1/19/2012 1

Background This section gives an overview of VDC and relates it to the broader use of technology in AEC practice. VDC builds on traditional (20 th century) practice Early in the 21 st century, the facility design-construction-operations process has many admirable properties. The AEC design-construction process creates the world s fixed physical wealth such as homes, offices, schools, power plants, and fixed systems of our lives, including water, waste, transportation and power distribution. However, the process has problems. The process is fragmented so that it takes a long time to complete projects (usually far too long for the owner; although not long enough for critics). The fixed wealth is expensive for all and increasingly so for the world s least advantaged. The US construction process has measurably declined in its productivity per human hour invested over the past forty years, although sister engineering fields have dramatically increased their productivity during this period [Teicholz 04]. From an immediate engineering perspective, the process has maddeningly long latency in the sense that it often takes days or even months to get information or decisions. Many project stakeholders feel effectively disenfranchised from the design-construction process because it is so arcane and complex. Finally, participants complain that it is paper based and inflexible. Figure 1 shows a photo of a construction planning office taken in 1998. We ask our students when the photo was taken. They guess the actual date fairly accurately and usually quickly recognize that it could have been taken in 1970 or yesterday except for some details. Some important methods of the construction planning process have changed during the past thirty years. For example, most of the paper documents shown in the photo are now printed computer documents, rather than copied hand generated documents. When asked if they expect an analogous photo to look similar thirty years from now, they uniformly say no! Interactive computer applications will replace most of the paper, and surely the roll of drawings on the table. One woman student suggested that there will be women in the picture. 1/19/2012 2

Figure 1: This photo was taken in a construction planning office in 1998. Some important characteristics of this process are unchanged in the past thirty years, e.g., there are multiple participants in the process, although probably many other important stakeholders are absent from the meeting. We suggest that most such photos will look differently in thirty years, and some today already do those from construction management projects that are significantly more efficient and effective than most. The big change, which already has started to occur, is the emerging use of interactive computer-based visual project models to improve communication of project information, reduce latency and increase collaborative ownership of project plans. Early History We introduced the term Virtual Design and Construction (VDC) in 2001 as part of the mission and methods of the Center for Integrated Facility Engineering (CIFE) at Stanford University, and we explicitly have used the VDC methods in research since that time, e.g., [Garcia 04]. Both of us use the VDC method in teaching and research, and one of us has taught a formal VDC class since 2001 [Kunz 05]. Most of our current and recent Ph.D. students use the method. Projects require developing and then relating design definitions, actual designs and design analyses, and then linking design, construction project management, with product management and financial management systems. In practice today, multiple teams perform most of this linking manually and socially, with great cost, interaction latency, and confusion. Motivated by these business drivers and technical work in concurrent engineering, which tried to integrate product, organization and process modeling and analysis tightly, the goal for interoperability emerged for multiple computer systems to exchange information and to use the exchanged information effectively. We now sharply distinguish social data exchange from technical or computer-based data exchange. DARPA supported Concurrent Engineering research that led to an integrated Product Organization Process model and design methodology [Londoño 89]. Londoño et al used a blackboard for communicating and for control of information flow. The domain concerned engineering parts, and the blackboard database described the Product, Process and Organization. As we use VDC in an Integrated Concurrent Engineering environment (see below), all of a project s stakeholders can access the integrated project database, and individual designers can modify and analyze details of current designs in local data 1/19/2012 3

spaces. Many researchers address complementary or similar issues. Prasad anticipated concurrent product and process organization and many of the related VDC issues that we discuss, e.g., in [Prasad 96a and b], and he discusses product and process organization. Industry Foundation Class (IFC) standards work of the International Alliance for Interoperability [IAI 05] and [Froese 02] both discuss semantic models for data exchange. More recently, the IFC and web XML data language and IFC communities collaborated to develop methods that provide web-based standards for sharing IFC data [aecxml 05]. The large Building Lifecycle Interoperable Software (BLIS) demonstration project defined hundreds of views and about 100 concepts, which were a practical subset of the IFC standard at the time [BLIS 02, 04]. The BLIS project was the first major demonstration by multiple software vendors to create an integrated set of project design models based on a shared IFC-based architectural model. The project had active participation from 1999 2002, and it demonstrated that real CAD and analysis applications could usefully share and exchange at least some of the data that a design team needed to create a design. While the BLIS project received lots of energy and attention, it had only limited success in its original hope to stimulate broad software developers support of IFCs and industry use of them. We have had mixed experience with use of computer application interoperability: it is possible to some extent, but difficult and limited [Kam 02], even in our own university teaching and research, as modest as they are. We find anecdotally that many AEC companies share similar concerns. Our VDC work commits to making explicit the semantics of data that practitioners of different perspectives and applications need to share, and attempting to facilitate practitioners to define and use shared explicit representations. Like the IAI effort, we encourage the project team to identify and commit to a standard vocabulary. Further, we recommend being inspired by standards such as the IFC and pragmatic like BLIS, but even more modest than both. However, unlike the IAI standards, we do not propose standards for the semantic details of VDC models and method; we pragmatically assume that the long-term solution will take a long time to emerge, and in the short term, we want to support individual project teams to do as well as they can with modest incremental effort. Further, we encourage a strict discipline on both the level of detail of VDC models and the process for creating whatever detail the project team wants, which we describe below in the section POP models have different levels of detail. Example VDC maturity model We find that users implement VDC in three distinct phases, each of which has its own value proposition, strategy for producing value and costs. Normally, organizations proceed sequentially through the steps in this maturity model, but some of the third step, Automation, often requires minimal and specialized, not general Integration. 1/19/2012 4

4. Visualization and Metrics: In this first phase, project teams create models of the Product in 3D, of the Organization that performs design, construction and operations and the Process followed by organizational participants to do design, construction and operations and management, based on performance metrics that are predicted from models and tracked in the process. The results of the CIFE- CURT VDC use survey find that this stage is in common (although not yet widespread) use within the global AEC industry [Kunz and Gilligan 07]. For Visualization to work well, all stakeholder organizations need to develop the competence to interpret the visual models, and many need to develop core competence to develop them, which requires a strategic investment in the methods and their use. Similarly, for Visualization to work well for multiple stakeholders, multi-party collaboration contracts need at least to allow and ideally to incentivize data sharing, which may require strategic change in partnering arrangements. In the Visualization phase, projects: Routinely model and visualize the most expensive elements of the Product, Organization and Process (POP); Use a social process among project stakeholders to integrate multiple VDC models and model versions; Justify investment in VDC tools, methods and human resources based on the value proposition to the project, since this phase is (relatively) inexpensive and individual projects receive can significant benefit; Clarify project objectives, values, responsibilities, designs and expectations because good visualization enables many more stakeholders to participate in project review far more meaningfully than in routine practice. 2. Integration (computer based): In this phase, projects develop computer-based automated methods to exchange data among disparate modeling and analysis applications reliably. Some vendors support data exchange among different applications using proprietary exchange methods, which often works well for those applications made by the particular vendor. The results of the CIFE-CURT VDC use survey find some evidence that some projects now use computer-based integration of two or more applications [Kunz and Gilligan 07]. For Integration to work well, vendors need to agree on exchange standards, which may require a strategic commitment to support crossvendor data exchange. Similarly, for Integration to work well for multiple stakeholders, multi-party collaboration contracts need at least to allow and ideally to incentivize data sharing, which may require strategic change in partnering arrangements. In the Integration phase, projects: Share data meaningfully among Product, Organization & Process models and analysis programs using interoperation, i.e., reliable computer-based data exchange. Cannot justify investment in VDC tools, methods and human resources based on their project value proposition. Rather, the value proposition must support the firm, since this phase is (relatively) expensive and multiple projects must use the same methods for the investment to produce significant benefit. 1/19/2012 5

IFCs are designed to enable this process, but there is little evidence that they are in significant use. Various vendors provide families of software applications that interoperate, often using proprietary exchange methods, which still limit exchange with other applications that might be useful to a project. Derive incremental value from integration per se because it can reduce modeling effort and time. 3. Automation In this phase, projects use automated methods to perform routine design tasks or to help build subassemblies in a factory. For Automation to improve design, project organizations normally need to dramatically change their processes to enable or perform more high-value design and analysis and spend much less time and billable effort for routine design. To support fabrication, the project needs to change from Design-Build or Design Bid Build to Design Fabricate Assemble, which takes strategic commitment to support a new partnering arrangement. Automation requires Integration, and good visualization helps make it work well. In the Integration phase, projects: Automate some aspects of routine design or Computer Numeric Control (CNC) manufacturing of assemblies for field installation Cannot justify investment in VDC tools, methods and human resources based on their project value proposition. Rather, the value proposition must support the firm, since this phase is (relatively) expensive and multiple projects must use the same methods for the investment to produce significant benefit. Enables dramatic increase in design efficiency and effectiveness; Enables dramatic decrease in construction duration, which in turn leads to breakthrough project performance in construction duration, e.g., the CIFE 2015 objective to be able to build most projects within 6 months from ground-break to high value use Themes VDC models are virtual The current practice of AEC design and construction, as shown in Figure 1, obviously works for developing value-adding projects today. Computer applications generated most of the paper documents shown in this photo, such as the project schedule on the wall, which is the output of a scheduling program, and the drawings, which are the output of a CAD program. Paper documents today provide high-resolution descriptions of project elements including architectural designs and plans, and the vast majority of AEC projects in use today were created using these paper-based methods. However, the discrete paper based documents do not help integration of different disciplines and making even simple changes requires hours to days to make the initial change, print and review the updated documents and do even simple updates to related documents of functionally related disciplines. In addition, the format of today s paper documents is often difficult for diverse stakeholders to understand: for example, users can rarely make meaningful comments about 2D architectural drawings or Gantt charts. 1/19/2012 6

Like their forbearers in practice today, VDC models are computer based. In addition, however, the use of the VDC project models is flexible, visual and interactive, not document or paper-based. The engineer who generates the VDC schedule can project it and show it to other stakeholders who have responsibility for the CAD model or some area of the design or construction. They in turn can project their CAD models or project digital photos, ideally simultaneously with the schedule, each on a separate projection screen such as that shown in Figure 2, a photo of a project meeting taken in our CIFE lab. Inexpensive computers and large, high definition and inexpensive projection devices enable social sharing of VDC computer models, and the modeling and simulation applications are now powerful and affordable. Figure 2: Photo taken in the CIFE lab of a construction planning meeting using Virtual Design and Construction methods. As in the traditional method of Figure 1, there are multiple stakeholders in the meeting. Models of the product, organization and process can be displayed, explained and updated simultaneously on the separate displays. We find that design team performance improves dramatically compared with in the traditional method of Figure 1, and our goal is to simultaneously improve project team performance in schedule, cost and quality dramatically. Interactive computer models replace traditional paper documents. The VDC model supports use by multiple stakeholders, as Figure 1 shows occurring in current practice and Figure 2 shows in the interactive VDC process. Since VDC is designed to support a multi-disciplinary project team, appropriate stakeholders include the multiple architects, engineers and general and multiple specialty contractors of AEC as well as owner representatives, users, suppliers, community representatives and government jurisdiction officials. VDC creates an integrated framework and set of methods to manage the project, including those aspects of the project that must and can be designed and managed, i.e., the building, the design-construction process and the organizations that follow the processes to design, build and use the building. Building Information Modeling (BIM) focuses on the building elements of the VDC model, which we find useful but limiting because management issues usually involve building organization process interactions. BIM definitely appears to hold promise in practice [Bedrick, 05], [Haymaker 05]. BIM today is enabling many AEC professionals to improve performance. However, 1/19/2012 7

even using best BIM practices, projects do not normally model, visualize or analyze the organization and process accurately and effectively, and methods to manage and communicate multidisciplinary information and processes remain ad-hoc. This paper includes a number of methodological recommendations on how to implement VDC in practice, which we enclose in boxes, such as the one below. Stakeholders can request the visual VDC models they need to participate effectively in the design, learn to understand all models as they evolve, and express their perspectives in a timely manner to other stakeholders throughout the project design. Suggestion: Invite all relevant stakeholders to the project kickoff meeting, including an owner representative, architect, major contractors, and a potential user. In the meeting, identify the VDC models and visualizations for the project that will help stakeholders provide meaning and timely input to the project design and management. Define the project product, organization and process vocabulary in a generic POP model as part of the kickoff meeting. Interactive VDC enables a very big change in the behavior of the design-construction process: dramatic reduction in decision latency, or the time between posing a question and having information with sufficient quality that it can be used to make a design decision. Questions can be formal Requests for Information or informal inquiries of fellow stakeholders. We repeatedly see latency change from days to hours and even minutes in integrated design sessions (see the section below on Integrated Concurrent Engineering). Natural visual VDC models make the content of each model much more accessible than they are in traditional static paper descriptions. Specifically, most stakeholders find that interactive 3D models are vastly more understandable than static 2D plan and section drawings, and 4D product-construction process animations are similarly much more understandable than traditional project schedule Gantt charts. Our interactive project models have started to become mutually parametric in the sense that change or highlighting any one will lead to very rapid or even instant change or highlighting in all others that are dependent. Because models can be examined with respect to each other, each grows to support the issues of others; time to get explanations and make decisions drops from days to seconds, likelihood of both design and construction rework drops because relevant stakeholders have increased ownership and timely participation in project decision-making. Since VDC models are visual, project team members who have different native languages can all reference the same graphic models, providing some support for the multi-cultural teams that are now common on many construction projects worldwide as well as some larger design teams. Some organizations have started to use multi-disciplinary VDC models as the focus of daily, weekly and major milestone design, planning and review sessions. Suggestion: Hold project kickoff, major review, weekly and daily design and construction project meetings in a room with multiple computers, ideally 1/19/2012 8

with at least three projected screens that all participants can see simultaneously. Plan the agenda around description, analysis and evaluation of product, organization and process issues as shown explicitly in models. Invite all relevant stakeholders to the project meetings. VDC models represent the Product, Organization and Process (POP) We set the broad goal to create explicit models of those aspects of a project that a manager can manage. A project manager can control three kinds of things: the design of the product to be built, the design of the organization that does the design and construction, and the design and the design-construction process that the organization follows. We call this project model the Product-Organization-Process model, or the POP model. The POP model is object-oriented in the sense that each P, O and P element has defined meaning (or semantics) to the stakeholders. For example, the Product model defines building elements such as Floors, Walls and Beams; the Organization model defines organizational groups, and the Process model defines activities and milestones. We define two related types of POP models: generic and instance. Generic models describe the conceptual vocabulary and thus can be very useful to define shared vocabulary for project stakeholders at the time of launching a project. More generally, generic POP models can define the vocabulary that a company or partnership uses to do a kind of work, allowing a community of organization professionals to define shared vocabulary that individual projects can customize as needed. Generic POP models define entity names, such as column, Design Team and activity. They may define associated attributes such as height, team responsibilities, and planned duration. However, generic models lack specific detail, i.e., have no values for their attributes, or names of individual instance elements. Instance models specialize the vocabulary as their generic relatives, naming individual elements, such as Design Team A. POP instances refer to corresponding objects in individual modeling or analysis applications, such as entities in a CAD, organization or process modeling application. They may refer explicitly to corresponding objects in modeling applications, or they may refer implicitly if POP model users understand the model naming correspondence. They may contain values of design variables such as planned dimensions when it is useful to share those values across multiple models. The POP model specifies information that is shared among models, not a complete project model with which individual modeling applications send and retrieve information. Thus, the POP model describes the content of the individual P, O and P models, each which represents the details of designed conceptual elements with their attributes, attribute values and relationships. The individual P, O and P modeling tools have user interfaces that present the models using natural visual idioms that are appropriate for each relevant discipline perspective. The purpose of the shared POP model is to define conceptual elements that are shared and help the stakeholders to assure that the product, 1/19/2012 9

organization and process specifications are appropriate and mutually consistent. For example, the product model defines the physical elements to be designed and built, at some selected but necessarily incomplete level of detail. The Organization model defines the groups that design and build each defined physical element, and the process model defines the activities and milestones that stakeholders follow to do their work. The product model, and hence the Product segment of the POP model, should represent the components and systems of the building. A well-designed POP model then also includes organization entities to design and build the systems and components, as well as the activities to do design and construction. The Industry Foundation Classes (IFC) define a large set of product components and systems such as Floors, Walls, Beams, and Equipment as well as some definition of organizational and process entities [IAI 05, Tarandi 03]. The Scope section of coarse level of detail POP models will include only a small subset of the total IFC specification. Projects that find it useful to define POP models at several levels of detail may represent more of the IFC specification at the finer levels of detail of their generic POP models. Suggestion: Whenever possible, create the vocabulary of the Form segment of POP models using the names and definitions of deliverable element (such as components, spaces and systems) of organization entities, or actors, and process activities as defined in the IFC specification. Doing so will facilitate making the resulting models IFC compliant. Create a cross reference between the product vocabulary based on the IFC specification and the terms used in product marketing and design. Similarly create a cross reference between the IFC-based organization and process vocabulary and the vocabulary that is in common operational use. At a high level, POP models represent the function, designed form or scope, and behavior of the project product, organization and process. Figure 3: VDC POP models represent the functional intent (Column 2), the designed form or scope, which the project team specifies, designs and builds in response to the functional intent (Column 3), and the predicted or measured behaviors of the critical project elements (Column 4). Project elements include the product, typically a facility (Row 2), the organization that designs, builds and operates the facility (Row 3), and the process that the organization follows as it does its project work (Row 4). Note that Building Information Models (BIM) represent the form/scope of the product, which is a crucial but small portion of both the total perspective of a project and the information about a project represented in the VDC framework and a POP model. Figure 3 schematically represents the overall content of generic POP models. Figure 4 shows a generic POP model in the sense that it specifies the vocabulary used to describe a line of business. Related instance models represent specific projects, such as that to create an individual building. 1/19/2012 10

Figure 3: VDC POP models represent the functional intent (Column 2), the designed form or scope, which the project team specifies, designs and builds in response to the functional intent (Column 3), and the predicted or measured behaviors of the critical project elements (Column 4). Project elements include the product, typically a facility (Row 2), the organization that designs, builds and operates the facility (Row 3), and the process that the organization follows as it does its project work (Row 4). Note that Building Information Models (BIM) represent the form/scope of the product, which is a crucial but small portion of both the total perspective of a project and the information about a project represented in the VDC framework and a POP model. 1/19/2012 11

Product Conformance to product objectives >= 99-2 Product Rentable area (ft2) range 300-400?p 2 Function Form/Scope Behavior Project Element Attribute Relationship Objective Choice Predicted Observed Assessed Product Product Scope Relationship Functional Requirement Product Scope (Space, System) Product Building Spaces include Offices Offices Product Building Spaces include conference rooms conference rooms Product Objectives Product Cost (K$) = 60?p 1 Product Energy (KBTU/sq-ft/year) <= 40?p -1 Organization Organization Scope Relationship Functional Requirement Organization Form (Actor) Organization Actors include Architect Architect Organization Objectives Conformance (Actor assignment to Organization Organization Function) (%) = 100 -?0 2 Organization Cost (K$) = 40?p?o 1 Organization Actor Backlog = 3?p?o 0 Process Process Scope (Task Action: Object) Actor Responsible Actor Process Form (Task Action: Object) Process Approve: design Actor Architect Approve: design Process Assess: Behaviors Actor Owner Assess: Behaviors Process Objectives Process Safety: lost work incidents = 0 -?o 2 Process Peak Quality Risk < 0.25?p?o -1 Project Evaluated goodness Sum: 8 Legend A-level model elements specification missing or needs to be assigned Predicted value that meets functional requirement Assessed value?o variable whose value is not yet observed?p variable whose value is not yet predicted?a variable whose value is not yet assessed Not applicable Figure 4: Level-A generic POP model for design and construction of buildings. This mostgeneric level of detail specifies that POP models represent the Product, Organization and Process in terms of functional requirements or design intent, Form or Scope or design choices, and behaviors or predictions and observations. Level A models (shown) have about one form/scope element each for product, organization and process; Level B have about ten, level C about a hundred, etc. As shown in Figure 4, we define the content of the POP models using the classic function form behavior taxonomy of design theory [Gero 90, Clayton 96]: Function, or design intent, represents the intent of the owner in making a requirement or the requirement of a critical stakeholder such as the code jurisdiction. Examples include that an auditorium seats 100; that an organization include a licensed structural engineer; and that the design process include certain specified review milestones. Form, or design choice or designed scope represents the decision of a designer in response to a functional requirement or the designer preference. Examples include choice of specific spaces, the choice of a particular contractual relationship among the architect and contractors, and the construction plan. 1/19/2012 12

Behavior, or properties, including both predicted behaviors of the design and measured behaviors of the product, organization or process. Examples include the predicted deflection of a beam; measured hours spent by a contractor doing a task; and predicted CPM duration of the construction. Figure 5 shows an example of a POP model that both specializes the generic model of Figure 4 for a university dormitory project and elaborates its level of detail. For example, the generic Level-A generic Product functions concern capacity and sustainability. At a greater level of detail, the instance Level-B Product functions include requirements to house 100 students, energy and water use that are stated fractions of 2002 comparables, specific noise, air quality open space, recycling and height objectives, the assumption of two-person dormitory apartments and a specific project budget. The generic Product form or scope represents systems and physical building elements, which remain unstated. As shown in Figure 5, the Level-B instances include ten two-bedroom apartments per floor at 400 square feet each and elements that still need to be sized including solar panels, a foundation pad, laundry, corridor, lab spaces, etc. In the judgment of the designers at this (still early) design stage, these physical elements represented the top ten physical elements in terms of cost. Their management goal is to design, procure and build these elements with predictable and acceptable costs, thereby minimizing the overall project cost risks. The team will elaborate the level of design detail once the design team has significant confidence that the design of these physical elements, the responsible design-construction team and the associated activities are all consistently specified and acceptable. The generic Product behaviors include the predicted and observed (by post occupancy measurement and evaluation) Product functions those listed in the function segment of the POP model and periodic assessment during the design and construction process of the conformance of the design to the design functional requirements. Many researchers use model-based computational methods to predict behavior, such as [Dym 88], [Shea and Cagan 99], Flemming and Woodbury 95] [Stiny 80]. Additional product behaviors include design team assessment of the conformance of the prediction of design performance to each stated design requirement and measured observations of the final building performance for each stated requirement, by appropriate post occupancy evaluation (POE). Similarly to the Product column of the instance POP model of Figure 5, the Organization and Process Functions elaborate and specify details of the corresponding generic functions of Figure 4. The Organization and Process form or Scopes represent those that the design team judged to represent the top-ten in terms of cost. Although the POP model does not represent them explicitly, an external organizational modeling software application would, and the explicit POP model helps the project team understand the relationships of the Product, Organization and Process Scopes to each other (e.g., organization actor responsibility for design and construction and specific Product elements, and relationship between process activities and Product elements). Finally, the team will measure the Level-B organization and process behaviors and use those results in management. 1/19/2012 13

Project Element Product Attribute Function Form/Scope Behavior Relationship Objective Choice Predicted Observed Assessed Product Conformance to product objectives >= 99-2 Product Rentable area (ft2) range 300-400?p 2 Product Scope Relationship Functional Requirement Product Scope (Space, System) Product Building Spaces include Offices Offices Product Building Spaces include conference rooms conference rooms Product Building Spaces include public areas public areas Product Building Systems include HVAC HVAC Product Building Systems include telecom/network telecom/network Product Building Physical Elements include foundation foundation Product Building Physical Elements include above-ground steel above-ground steel Product Building Physical Elements include drywall drywall Product Building Physical Elements include skin skin Product Building Physical Elements include windows windows Product Building Physical Elements include roof roof Product Objectives Product Cost (K$) = 60?p 1 Product Energy (KBTU/sq-ft/year) <= 40?p -1 Organization Organization Scope Relationship Functional Requirement Organization Form (Actor) Organization Actors include Architect Architect Organization Actors include City City Organization Actors include Concrete sub Concrete sub Organization Actors include Flooring sub Flooring sub Organization Actors include GC GC Organization Actors include MEP sub MEP sub Organization Actors include Owner Actors Organization Actors include Painters Painters Organization Actors include Steel sub Steel sub Organization Actors include Structural Engineer Structural Engineer Organization Objectives Conformance (Actor assignment to Organization Organization Function) (%) = 100 -?0 2 Organization Cost (K$) = 40?p?o 1 Organization Actor Backlog = 3?p?o 0 Process Process Scope (Task Action: Object) Actor Responsible Actor Process Form (Task Action: Object) Process Approve: design Actor Architect Approve: design Process Assess: Behaviors Actor Owner Assess: Behaviors Process Design: Building elements Actor Architect Design: Building elements Process Design: Building systems Actor HVAC/MEP designers Design: Building systems Process Predict: Predictable Behaviors Actor Owner Predict: Predictable Behaviors Process Build: Building elements Actor GC Build: Building elements Process Build: Building elements Actor Flooring sub Build: Building elements Process Build: Building elements Actor GC Build: Building elements Process Build: concrete elements Actor Concrete sub Build: concrete elements Process Build: Flooring Actor Flooring sub Build: Flooring Process Objectives Process Safety: lost work incidents = 0 -?o 2 Process Peak Quality Risk < 0.25?p?o -1 Conformance (Actual schedule to Process plan) (%) > 80-76 0 Process Peak Predicted Schedule Risk (wks) < 2 -?o 1 Figure 5: an instance Level-B POP model represents the Function (intent), Form/Scope (design choices) and Behavior (properties) of a project Product, Organization and Process. The broad goals of the POP model are to help the stakeholder team to identify major requirements, the most expensive design choices made by the design team to meet those requirements, and the predictable and observable project behaviors early in the design process. The hope, and our experience, is that the POP model helps enable the most valuable possible modeling and analysis of a project during its entire lifecycle. POP models have different levels of detail 1/19/2012 14

Arbitrarily, we define POP model level of detail as a power of ten. For consistency with traditional product, organization and work breakdown structures (see Use Breakdown Structures to define generic POP models below), a Level-A POP model represents the product, organization and process as a single element, e.g., the building, designconstruction team and design-construction process. Useful as a reference, this level of detail is too abstract to have managerial interest. A Level-B POP model represents P, O and P elements that each incur about 10% of the project cost, design-construction effort or schedule duration. Figure 5 is an example. This initial Level-B POP model shows the P, O and P design elements that, in the judgment of the project team, represented the elements that will require the greatest cost, effort or schedule at this level of detail. The broad objectives of the POP model are to help the stakeholder team to identify these resource users explicitly early in the design process and to enable consistent modeling of those elements in the associated product, organization and process models. Having developed such a model and understanding its significance for managing the project, we can then elaborate its detail to a Level-C POP model, which represents those POP elements with about 1% of the cost, effort or duration. While AEC projects often define Level-D or greater level of detail, we focus our research and this paper on Level-A to C models. We borrow the convention of ABC from the method of Pareto analysis, or ABC analysis. With respect to the total cost, schedule or quality, we try to guild POP models that identify the ten most important factors in the A category. We try to identify the ten or so most important sub factors of each A category as the hundred factors in the B category, and in turn we will identify the ten sub factors of each B category as members of the relatively unimportant C category. A measure of success or goodness of a POP model is that, for a particular level of detail, its elements are mutually consistent in detail, mutually refer to each other, and together describe the most important aspects of a project at a particular level of detail. For example, a Product element should relate to a design and a construction activity and to organizational parties (called actors ) responsible for those activities. Similarly, a good POP model represents the product components with each modeled actor and activity related. Our early simple versions of the POP model do not explicitly represent the relationships among P, O and P elements, although the stakeholder team knows them. The greatest value of the POP model comes when the project influence is greatest, i.e., at the schematic and early design development phases. At these early phases, by definition, only Level-A or B and possibly C details Figure 7 shows a set of guidelines for how to create individual functions (design intent), form/scope (design choices) and behaviors (parameters) of a project. The modeling purpose might be to support understanding of architectural concepts, space management (i.e., how much space is planned and actually available for different spatial functions), cost estimation, energy analysis, schedule optimization, schedule impacts on stakeholders, structural analysis or mechanical system design and analysis. The choice of 1/19/2012 15

particular product, organization and process forms will depend on the choice of modeling purpose. We recommend the following process to build and refine POP models: Build a POP model as a stakeholder team at a particular level of detail (first A, then B, etc.), iterating until the team agrees on the consistency of the function, form and behavior of the product, organization and process o Set functions (objectives) of Product, Organization and Process; o Design the form or scope of Product, Organization and Process in response to the functional objectives; o Identify project behaviors to predict or measure and the relative importance of each to overall project functional objectives, e.g., by distributing 100 points among all behaviors; define methods to predict, assess and observe them. o Identify threshold values that define different levels of goodness of each predicted or measured behavior value, e.g., the range of values that define values for very good, good acceptable and unacceptable. Based on the POP model, build detailed product, organization and process models and predict objective behaviors using model-based or other methods. o Based on result of predictions or professional judgment, add or revise values of predicted or observed P, O and P Behaviors; o Assess the goodness of each behavior with respect to the goodness thresholds, e.g., on a 4-point scale: 0, 1, 2, 3, which the POP model can do automatically; o Assess overall goodness for project and compare relative design goodness of this option vs. others. The Analyses tab of the CIFE POP model implements the decision-analysis framework described in the MACDADI project [Haymaker] in which overall design option goodness is defined as the sum of the assessed goodness of each behavior weighted by its relative importance. See the simple example of Figure 6. o Evaluate the acceptability of the behaviors of a model option, given its forms and functions: identify the option that has highest relative goodness (e.g., from Center in Figure 6) and, if it still does not meet project requirements, identify objectives that have lowest assessed relative goodness for one or more design options of interest (e.g., from Left of Figure 6). o Refine the model: assign management and then team attention to refine the model or the analyses at the current level of detail. Once the POP model is acceptable at a particular level of detail, then the team can choose specific areas of risk or opportunity and assign attention to refining the model. To elaborate POP model details to the next level of detail, e.g., Level-A to B or Level-B to C. 1/19/2012 16

Figure 6: Three views of the assessed goodness of a project, which is an output of the POP model: (left) relative goodness of each of the defined project objectives for two design options, (middle) overall assessed project goodness of each option, which is the weighted sum of the objective goodness of each project metric, and (right) a business summary of the project cost and evaluated goodness of the options as designed, given the functions and form choices. Use Breakdown Structures to define generic POP models POP models represent three facets of the project, and the models start with generic Product, organization and Process (Work) Breakdown structures. A goal of a project POP model is to define the PBS, OBS and WBS so that they represent the important characteristic of their respective project models and are mutually consistent in both naming and references. The scopes of each row of a POP model should be consistent with the corresponding PBS, OBS or WBS. Suggestion: Create a Level-B instance POP model very early in the design process, at least by the end of the first day of a kickoff meeting. Start to elaborate the level of detail to Level-C only after the Level-B elements are defined, modeled, mutually consistent, acceptable and well understood. Suggestion: Choose Level-B POP model elements so that each represents about 10% of the cost, effort or schedule duration of the project. That is, model the physical product components, organizational actors and activities that represent about 10% of the cost, effort or schedule duration of the project, whichever is most important for project success. The POP model represents breakdown structures hierarchically. Figure 8 Figure 10 show a generic PBS, OBS and WBS, which represent a hierarchical decomposition of types of product elements, organization elements and work respectively. The individual product, organization and process models will use the names defined within the breakdown structures and shown together in the Form segments of the POP model. The relationships in each breakdown structure represent a class-subclass specialization hierarchy. That is, Product, Organization or Work elements at the top of each BS are abstract; lower levels become increasingly more specific and specialized. Suggestion: Include all the relevant stakeholders in defining breakdown structures and each major version of the POP model, at least including the owner representative, architect, contractor, and a potential user. 1/19/2012 17

The PBS represents the physical and abstract components that together represent the physical and functional facility being built. The PBS represents both physical components to be designed and built, such as columns and slabs, as well as abstract systems such as egress and ventilation systems. The PBS shown in Figure 8 is based on the Industry Foundation Classes (IFC) specification [IAI 05, Tarandi 03]. In the simplest generic POP models, entities have only name; there is no explicit representation of their attributes. Generic POP models can also define the names of the most important attributes of each entity type. The relationships among P, O and P entities should be clear to the stakeholders, but they are normally implicit in simple POP models, as shown in Figure 7. 1/19/2012 18

Product Organization Process Function Scope Behavior Measurable required functional project capability that is OK at each significant project milestone: qualitative or quantitative, e.g., specific spaces and specific systems, energy, lighting and egress, capacities and performance of spaces and systems, budget, assessed by responsible stakeholders Design Intent Intended skills and responsibilities of the project actors who are stakeholders Measured actor meeting participation timeliness: high participation of intended stakeholders > 90% of meeting possibilities Major milestone dates, including start and finish VDC modeling purpose and Level of Detail Activity schedule conformance: specific value and variance objective as well as measured values and variance Activity budget conformance: objective and variance Allowed predicted actor backlog, cost Design Choices Physical spaces, components and Actors to achieve Product, Activities to achieve Product, systems; abstract deliverables to Organization and Process Organization and Process achieve Product functional functional objectives functional objectives objectives R: designed scope R: designed form or R: each form implements one implements functions scope implements or more product functions Each actor has functions Each scope represents about responsibility for about Each activity represents 10% of the project TCE 2 10% of the project TCE 2 about 10% of the project (Level-B) (Level-B) TCE 2 (Level-B) R: each physical element has R: actor has assigned process Task(s) activities Properties: analysis predictions and observed performance Measurement quality of designed or Predicted, measured actor Predicted risks, measured delivered product assessed by R: Behaviors have schedule delay responsible stakeholders quantitative objectives R: Behaviors have R: Behaviors have quantitative stated in functional quantitative objectives objectives stated in functional requirements stated in functional requirements requirements Predicted, measured actor backlog Measured architectural, construction, energy, etc. quality conformance Predicted, measured organization costs Predicted, measured direct and hidden work volume Measured direct and hidden work volume conformance Measured actor meeting participation timeliness Predicted, measured schedule conformance Measured Process cost conformance Figure 7: This table shows a set of guidelines for how to create individual Functions, Form/Scopes and Behaviors and how they appropriately relate to each other. Analysts can predict or measure behaviors by day, week, month, major milestone, etc. as appropriate. R: indicates relationships among POP elements, which normally are required for a consistent POP and project model, although they are implicit in the model itself. For example, individual Product Form or Scopes specify spaces, components and systems, which the design team chose individually and collectively to satisfy product Functional objectives. TCE 2 is the total predicted Time, Cost, Effort or life cycle Energy use. 1/19/2012 19